zlacker

>>isaacf+(OP)
OpenAI isn't as open as it used to be, but they still publish papers.

The open-source Hadoop was based on Google papers on MapReduce and Google File System.

GPT-J and GPT-NeoX are open source models based on OpenAI publications around GPT-3.

>>jarrel+sl1
Even Apple publishes ML papers. It turns out you can't attract and retain decent research talent unless you allow them to continue racking up their google scholar score

>>sharkj+Sn1
Yes, publishing papers is good. Open source code can be based on methods described in papers. Like, most algorithms in sklearn for example.